Projects
Things I've built under PookieB — across web, iOS, and AI/ML.
Shredders
Real-time powder conditions for 53 ski resorts across 11 regions. 90 API routes, 7 data sources, 8-factor powder scoring, AI chat, social trip planning. Next.js, Supabase, SwiftUI.
SPACEc
Stanford academic Python library for multiplexed imaging analysis. 19,123 lines covering the full pipeline from cell segmentation to spatial analysis, with a novel Patch Proximity Analysis method. Published on PyPI and bioRxiv.
Also Built
AI financial decision engine with a 3-stage pipeline: natural language question → deterministic calculations → narrative explanation. 13 AWS Lambda functions orchestrated by Step Functions, Bedrock Claude for analysis, GPT-4o Vision for receipt OCR. SwiftUI companion app for expense splitting.
AI language tutor with conversation practice across 5 languages, FSRS spaced repetition, journal with real-time AI analysis, and a persistent learner memory that adapts to your mistakes and interests. Next.js, Supabase, dual AI providers.
ATC radio trainer for student pilots. 20 scenarios across 4 categories, ICAO phraseology scoring (0–100), voice I/O, live METAR/TAF weather. Built for learning to talk to air traffic control.
Automated news aggregation from 9 sources with AI-generated summaries and NPR-style podcast episodes via custom ElevenLabs voice. iOS app with widgets.
Social journaling with photo uploads, streaks, accountability partners, and trip organization with MapKit on iOS.
iOS relationship coaching app with AI coach, custom habit trackers, partner pairing, communication analysis, and WidgetKit integration. 86 Swift files.
Automates 17 ADU permit eligibility checks against San Jose regulations using 11 government GIS APIs. FEMA flood zones, fire severity, zoning, building footprints.